SIP Connection - Deployment Issues

Deployment Issues

If the call traffic runs on the same connection with other traffic, such as email or Web browsing, voice and even signaling packets may be dropped and the voice stream may be interrupted.

To mitigate this, many companies split voice and data between two separate internet connections. Alternately, some networks use the TOS precedence or DiffServ fields in the header of IPV4 packets to mark the relative time-sensitivity of SIP and RTP as compared to web, email, video and other types of IP traffic. This precedence marking method requires that all routers in the SIP and RTP paths support separate queues for different traffic types. Other options to control delay and loss include incorporating multiple VLANs (virtual local area networks), traffic shaping to avoid this resource conflict, but the efficacy of this solution is dependent on the number of packets dropped between the Internet and the PBX. Registration is required if the end user has a dynamic IP address, if the provider does not support static hostnames, or if NAT is used. In order to share several DID numbers on the same registration, the IETF has defined additional headers (for example "P-Preferred-Identity", see RFC 3325). This avoids multiple registrations from one PBX to the same provider. Using this method the PBX can indicate what identity should be presented to the Called party and what identity should be used for authenticating the call. This feature is also useful when the PBX redirects an incoming call to a PSTN number, for example a cell phone, to preserve the original Caller ID.

Users should also be aware that a SIP connection can be used as a channel for attacking the company's internal networks, similar to Web and Email attacks. Users should consider installing appropriate security mechanisms to prevent malicious attacks.